Sustainability and Quality in Wine Pricing: Insights from the Primitivo E-Commerce Market

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

I have gone through the article "Sustainability and Quality in Wine Pricing: Insights from the Primitivo E-commerce Market". This is an interesting topic that is closely related to the scope of Sustainability. However, the manuscript should be improved in the following areas.

Abstract

(1) The research background seems rather weak and fails to convey the urgency of "sustainability".

(2) The title mentions both sustainability and quality. Therefore, it is necessary to clarify the relationship between sustainability and quality, as well as the motivation for studying these two issues.

Introduction

(1) The author must point out the problems encountered by enterprises in the real world.

(2) Authors must state existing literature on such real-world problems。

(3) Please improve the introduction to be clearly stated research questions and targets first.

(4) The authors must still emphasize the originality and the motivations of the study.

Literature Review

(1) Does the paper demonstrate an adequate understanding of the relevant literature in the field and cite an appropriate range of literature sources? Is any significant work ignored?

(2) During the last 5 to 10 years, we have published several papers that may help you strengthen your arguments and deepen your analyses.

(3) This section is necessary for you to clarify the "contribution" of your study.

(4) To distinguish this paper from the existing literature, it is important to establish comparative table that explicitly shows the main differences between the paper and the relevant existing literature in the literature review section.

Methodology

(1) Please have the author better explain how this method is applicable to solving the problems raised in this research.

Conclusion

(1) Please make sure your conclusions' section underscores the scientific value-added of your paper, hardly allows the reader to appreciate the value of their results.

Author Response

We would like to sincerely thank the reviewer for the careful reading of our manuscript and for the constructive comments and suggestions provided. We have carefully addressed each point raised, as detailed in the responses below, and have revised the manuscript accordingly. We believe these changes have improved the clarity and overall quality of the paper.

Abstract

Comment 1: The research background seems rather weak and fails to convey the urgency of "sustainability".

Response 1: We agree that the previous abstract did not sufficiently emphasize the urgency of sustainability in the wine sector. Accordingly, we have revised the opening to explicitly highlight the pressing challenges that the sector is facing, including climate change, resource scarcity, rising energy costs, and shifting consumption patterns. In this way, the abstract now conveys sustainability not only as a consumer preference but also as a strategic lever of competitiveness and resilience in response to these urgent issues.

“The wine sector faces urgent challenges related to climate change, resource constraints, rising energy costs, and shifting consumption patterns. In this context, sustainability represents a lever of competitiveness and resilience, enabling producers to differentiate themselves through their connection with the territory and the preservation of artisanal practices.”

Comment 2: The title mentions both sustainability and quality. Therefore, it is necessary to clarify the relationship between sustainability and quality, as well as the motivation for studying these two issues.

Response 2: In the revised version, we have clarified that the repositioning of native varieties, such as the Apulian Primitivo, is based on production choices that combine sustainability and craftsmanship. These practices simultaneously safeguard environmental, economic, and social dimensions and generate wines with distinctive organoleptic qualities. In this way, the abstract now makes explicit that sustainability-oriented practices also contribute to quality, thereby justifying the joint focus on these two issues.

“This is particularly evident for native varieties, such as the Apulian Primitivo, whose market repositioning is based on production choices that combine sustainability and craftsmanship, generating wines with distinctive organoleptic characteristics and while also contributing to the environmental, economic, and social protection of the production areas.”

Introduction 

Comment 1:  The author must point out the encountered by enterprises in the real world

Response 1: We have added a section that explicitly addresses the difficulties faced by enterprises in the real world when implementing sustainability. The revised text now describes the most recurrent barriers for small and medium-sized enterprises—such as lack of resources, high initial capital costs, and limited managerial and technical skills (Álvarez Jaramillo et al., 2019). We also highlight the specific challenges encountered in the agricultural sector, such as the lack of favourable policies and the lack of skilled labor (Restrepo Campuzano et al., 2023).

Section 1.Introduction, p. 2, lines 46-59: Borsellino et al. (2020) [2] underline that the transformation underway represents both a challenge and an opportunity: creating effective, inclusive, and nutrient-conscious marketplaces that can reduce environmental externalities, promote collaborative governance models throughout the supply chain, and value low-impact agricultural approaches (integrated, organic, and precision farming) is essential. However, the literature shows that the transition to sustainable production models encounters numerous barriers, which vary according to firm size, sector, and adopted technologies. In the case of small and medium-sized enterprises, for example, the most recurrent constraints concern the lack of resources, high initial capital costs, and the lack of managerial and technical skills [3] (Álvarez Jaramillo et al., 2019). The systematic review by Campuzano et al. (2023) [4], which focuses precisely on identifying barriers to sustainable innovation in the agricultural sector, highlights the persistence of both internal (financial resources, skills, organizational mindset) and external barriers, including the lack of favorable policies and regulations and the scarcity of skilled labor.

Comment 2: Authors must state existing literature on such real-world problems。

Response 2: We have expanded the literature review to include studies that document these real-world barriers across different contexts. In particular, we now reference Álvarez Jaramillo et al. (2019) on SMEs and Restrepo Campuzano et al. (2023) on agricultural constraints. In addition, we provide an applied example from the wine sector, where several studies show both the structural difficulties of the transition (Wagner et al., 2023; Golicic, 2022) and the opportunities demonstrated by empirical assessments of sustainable practices (Falcone, 2016; Casson, 2022; Boyd, 2025). This reinforces the connection between the challenges faced by enterprises and the existing body of literature.

Section 1. Introduction, p. 3, lines 102-110: At the same time, several studies have shown that the transition from conventional practices to lower-impact solutions can, under specific soil and climate conditions, strengthen economic margins by combining yield stability with improved product quality. In particular, research conducted in Italian wine contexts has shown that the adoption of integrated approaches makes it possible to combine environmental and economic sustainability, albeit with some trade-offs related to yields [13]. Similarly, low-input regimes increase operational complexity and short-term risks but, under specific yield profiles and pedoclimatic conditions, allow for the joint achievement of environmental and economic performance. [14,15].

Comment 3: Please improve the introduction to be clearly stated research questions and targets first.

Response 3: In the revised version of the manuscript, the introduction has been improved to clearly present the research gap, the aims of the study, and the specific research questions. We now explicitly state two research questions (RQ1 and RQ2) at the end of the introduction, together with the description of the hedonic pricing model as the methodological approach adopted to address them. This revision makes the targets and scope of the paper clearer and more transparent to the reader.

Section 1.3 Research objectives, pp. 6-7, lines 298-304: Considering this evidence, the present study aims to answer the following research questions:

RQ1. To what extent do artisanal and sustainable practices (e.g., low-yield farming systems, use of indigenous yeasts, environmental sustainability mentions, and Geographical Indications) affect the price of Primitivo wine in the e-commerce market?

RQ2. What is the relative weight of these practices in the different market segments, from the mass market to the premium market?

Comment 4: The authors must still emphasize the originality and the motivations of the study.

Response 4: In the revised introduction, we have clarified both the motivations of the study, highlighting the literature gaps on sustainable practices and native varieties, the opportunities offered by e-commerce, and the relevance of the Apulian Primitivo context, and the originality of our approach, which lies in a threefold contribution: (i) theoretical, by extending hedonic models to sustainability-related attributes; (ii) methodological, through the application of quantile regression in an e-commerce setting; and (iii) practical, by guiding producers and policy makers in supporting native varieties and the sustainable transition of the wine sector.

Section 1.3 Research objectives, p. 7, lines 310-318: This approach delivers a threefold contribution to the literature: theoretical, by extending wine hedonic models to incorporate previously unexplored sustainability dimensions; methodological, through the application of quantile regression in this field; and practical, because quantifying the economic contribution of individual sustainability and quality practices to the formation of the final price, can provide strategic guidance to producers on investments, denomination policies, and digital communication, as well as policy recommendations for valorising native varieties and supporting the sustainable transition of the entire supply chain.

Literature review

Comment 1: Does the paper demonstrate an adequate understanding of the relevant literature in the field and cite an appropriate range of literature sources? Is any significant work ignored?

Response 1: We have nevertheless carefully checked the literature review to ensure that it provides a comprehensive coverage of the relevant works, and we also expanded the references discussed throughout the manuscript to strengthen the theoretical framework and to better position our contribution within the existing literature.

Comment 2: During the last 5 to 10 years, we have published several papers that may help you strengthen your arguments and deepen your analyses.

Response 2: We appreciate this remark and have carefully reviewed our literature section to ensure that it provides a comprehensive and balanced overview of the field. We are confident that the key contributions have been considered and integrated into our discussion.

Comment 3: This section is necessary for you to clarify the "contribution" of your study.

Response 3: To better highlight the contribution of our work, we added the following sentence at the end of the literature review (Section 1.2 Literature review, p. 66, lines 268–270):

“This overlooked dimension offers a promising avenue for further investigation, which the present study aims to address.”

This addition is intended to clarify more explicitly the positioning of our study with respect to the existing literature, while avoiding overlaps with the subsequent section on research objectives.

Comment 4: To distinguish this paper from the existing literature, it is important to establish comparative table that explicitly shows the main differences between the paper and the relevant existing literature in the literature review section.

Response 4: In our manuscript, we aimed to integrate the comparison with previous works directly within the narrative of the literature review and the research objectives, so as to maintain a coherent flow. Nevertheless, if the editors consider that a comparative table would add clarity for the reader, we would be glad to prepare and include it in the revised version.

Methodology

Comment 1: Please have the author better explain how this method is applicable to solving the problems raised in this research.

Response 1:  We have clarified the rationale for using quantile regression in two sections of the revised manuscript:

• in the Research objectives (Section 1.3, p.6, lines 306–310), where we highlight that quantile regression represents a methodological innovation in the e-commerce context, as it allows us to analyse not only average market effects but also the distributional tails, while taking advantage of standardized digital marketplace information;
• in Specifying the Quantile Regression Model (Section 2.2, p. 10, lines 438–444), where we further explain, following Cacchiarelli et al. (2014), that the extremes of the price distribution in the wine market should be considered meaningful rather than outliers, and that quantile regression is therefore more appropriate than transformations that compress these values.

These clarifications aim to make more explicit how the method directly addresses the problems raised in our research.

Conclusion

Comment1: Please make sure your conclusions' section underscores the scientific value-added of your paper, hardly allows the reader to appreciate the value of their results.

Response 1: 

We thank the reviewer for this valuable remark. In the revised version, we strengthened the Conclusions section to better highlight the scientific contribution of the study. In particular, we added the following passage (Section 5. Conclusion, p. 20, lines 739-746):

“To conclude, this work contributes to expanding hedonic models on wine by introducing variables related to sustainability and production identity, and it demonstrates the usefulness of quantile regression to analyse the heterogeneity of the online market. More generally, the results also offer insights into the role of the digital economy in sustainable development, showing how the greater transparency of information in e-commerce channels can affect the enhancement of production practices and, indirectly, purchasing behaviour.”

This addition aims to make clearer the theoretical, methodological, and empirical value-added of our results.

Reviewer 2 Report

Comments and Suggestions for Authors

Dear Author,

 

I found your article very interesting, particularly the recommendations for action and the outlook for future studies, as well as the self-critical comments regarding the limitations of the statements.

I have a few points of criticism, which are not intended to detract from the quality of your article, but rather to highlight the limitations of the model presented in the text:

1. It should be noted critically that the ‘hedonic pricing’ model cited by Rosen (reference no. 31) is based on a number of assumptions (such as complete information and market identification) that may not readily be available in the field of online wine sales. Furthermore, the specific dummy variables used for sustainability practices have not yet been certified, i.e. their status has not yet been verified (in consultation with consumers).
2. The numerous dummy variables must also be critically examined with regard to their correlation, i.e. how strongly these independent variables correlate with each other, even if the use of simultaneous quantile regression allows for a differentiated analysis along the price distribution (Q10–Q90) and describes the different dispersion of the data points around the regression line more robustly than OLS. However, questions remain about the sensitivity of the estimates to sample size and bootstrap parameters.
3. The log transformation of the price (ln_PriceL) is frequently used, but it would also be useful to check the robustness with alternative specifications, as although semi-log models are common, other forms (such as mixed-log) could potentially enable a better model fit.
4. Brand reputation beyond user reviews (e.g. by renowned and recognised wine critics) is of course not taken into account here.
5. A cross-sectional study is, of course, only a snapshot and unfortunately does not allow any conclusions to be drawn about changes over time in the ‘willingness to pay’ for sustainability and identity practices.

 

The graphics in Figures 1 and 2 are legible, but unfortunately they are low resolution and axis labels are missing (units of the x and y axes in Figure 1 and units in Figure 2 on the x-axis). The legends are very concise, meaning that experts in the field will understand them, but new readers will need to familiarise themselves with them.

I would have liked to read a little more detail about the software used, ‘STATA 15’ (command structures, setup, etc.).

When collecting data, I would also have liked to learn more about the source code of the scraper, or the libraries used (e.g. Python packages) as well as a description of any data cleansing that was carried out (if necessary).

The standard errors of the regression were estimated using ‘bootstrap’, but I could not find any information on the number of iterations, the confidence level or the selection of random samples, which at least makes reproducibility difficult.

Of course, it would be wonderful if the analyses could be verified via supplementary data packages or an open-source repository for the data sets and analysis scripts. Perhaps you should include a reference stating that interested readers can obtain this from you on request?

Author Response

We are very grateful to Reviewer  for the time and effort dedicated to evaluating our manuscript. The insightful remarks and recommendations have been extremely helpful in refining the analyzes and strengthening the discussion. All comments have been addressed in detail below, and the corresponding revisions are highlighted in the revised version of the manuscript.

Comment 1: It should be noted critically that the ‘hedonic pricing’ model cited by Rosen (reference no. 31) is based on a number of assumptions (such as complete information and market identification) that may not readily be available in the field of online wine sales. Furthermore, the specific dummy variables used for sustainability practices have not yet been certified, i.e. their status has not yet been verified (in consultation with consumers).

Response 1: 

As already discussed in the manuscript (Section 2. Material and methods, p. 7, lines 328–345), we explicitly acknowledge the strong assumptions underlying Rosen’s framework and the related econometric limitations: “However, this theoretical framework is based on robust assumptions: (i) perfect identification of the relevant market; (ii) independence of attributes or related implicit marginal prices; (iii) perfectly competitive markets in long-term equilibrium; (iv) perfect and symmetrical information between consumers and producers [29].

In the wine-growing context , Miller et al. (2007) [44] highlight how these conditions are difficult to achieve. Furthermore, as pointed out by Unwin (1999) [29], Rosen's framework [42], does not provide operational indications on the functional specification of the price function nor on the selection of the relevant attributes, exposing the analysis to econometric criticalities such as multicollinearity, endogeneity and sample representativeness problems. .

Despite these limitations, HPM has established itself as a reference methodology in wine sector studies, by virtue of its ability to estimate the economic value of quality and sustainability attributes incorporated in wines [45]. The literature reports a wide variety of methodological approaches used for the estimation of hedonic price functions. Ordinary least squares (OLS) regression is the most widely used technique, used in 93% of studies and 85% of estimates [28], thanks to the simplicity of interpretation of the coefficients and the availability of correctives for heteroskedasticity and autocorrelation (51% of studies).”

We also justify the choice of focusing on specialized online platforms (Section 2.1 Data collection and dataset construction, p. 8, lines 384–394), which provide standardized information and reduce asymmetries: “The decision to focus on these specialized platforms responds to a twofold need. On the one hand, these are direct sales channels that play an important role in the Italian market. According to the most recent Mediobanca report [50], pure players represent a market share close to that of producers' direct sites, with an overall sales growth of +8% in 2024 compared to the previous year. Tannico confirmed its position as market leader (€29 million in revenues), followed by Bernabei (€28.5 million), Vino.com (€27 million), Callmewine (€13.6 million), and Xtrawine (€9.1 million). On the other hand, the choice of the online channel allows access to detailed and standardized product information, reducing informative asymmetries and allowing a homogeneous detection of the technical and commercial characteristics of the wines”.

 

In addition, we have revised the section Limitations and future research to clarify that some dummy variables referring to sustainability practices are not yet certified nor validated through consumer consultation. Accordingly, the estimated values should be regarded as preliminary, and we suggest that future research could rely on stated preference methods (e.g. discrete choice experiments, willingness-to-pay analysis) to verify and quantify more accurately the value consumers attribute to such practices (p. 17, lines 712-719): “Finally, although the hedonic approach is useful in analysing the relationship between observable attributes and price, it does not allow for direct investigation of individual preferences and cognitive processes underlying purchasing decisions, especially considering that some sustainability variables introduced as dummies have not yet been certified or validated through direct comparison with consumers. In this regard, integrating stated preference method, such as choice experiments or willingness-to-pay (WTP) analysis, could provide a more comprehensive understanding of the value consumers attribute to different production and communication practices.”

Comment 2: The numerous dummy variables must also be critically examined with regard to their correlation, i.e. how strongly these independent variables correlate with each other, even if the use of simultaneous quantile regression allows for a differentiated analysis along the price distribution (Q10–Q90) and describes the different dispersion of the data points around the regression line more robustly than OLS. However, questions remain about the sensitivity of the estimates to sample size and bootstrap parameters.

Response 2:  We examined the correlation matrix between the dummy variables included in the model. The analysis revealed moderate correlations between Score_B and Score_C (0.68), and between more_75cl and Pack_ast_cass (0.66). To reduce the risk of multicollinearity, we made some changes to the model specification. First, we changed the reference category for the score-related variables: Score_B was set as the baseline, while Score_A and Score_C, which were uncorrelated, were retained in the regression.

Second, regarding the correlation between more_75cl and Pack_ast_cass, we believe it is theoretically difficult to exclude one of the two variables, as they capture distinct but related aspects of product presentation: more_75cl reflects the bottle format, while Pack_ast_cass represents a premium packaging strategy (wooden cases and crates) that may accompany, but does not necessarily coincide with, larger formats. To reduce collinearity, we reformulated the price variable in terms of "price per 75cl," thus eliminating the format from the explanatory variables and retaining the packaging-related variables, which are relevant for estimating its impact on price. These measures reduced the risk of bias due to multicollinearity while simultaneously preserving the interpretive value of the variables of interest.

To test the sensitivity of the estimates to the bootstrap parameters, we repeated the simultaneous quantile regressions with 200, 500, and 1000 replicates. The coefficients were identical, while the significance values ​​showed slight variations without changing the classification of the variables with respect to the conventional levels (n.s., *, **, ***). This confirms the robustness of the estimates to the choice of the number of bootstrap replicates. In addition, the Wald test, which will be included in Appendix A, confirms that some attributes (e.g. designation of origin, alcoholic strength, environmental practices, indigenous yeasts and sapling training) present significant heterogeneity along the conditional distribution of prices.

Comment 3: The log transformation of the price (ln_PriceL) is frequently used, but it would also be useful to check the robustness with alternative specifications, as although semi-log models are common, other forms (such as mixed-log) could potentially enable a better model fit.

Response 3: 

Functional specification by the Box–Cox test suggested a value of λ ≈ –0.71. However, the corresponding transformation implies an excessive compression of the variability at the extremes of the distribution (Table X), with the risk of distortion precisely in the upper tail, which is central to our analysis. As Cacchiarelli et al. (2014) point out, in the wine market, the extremes of the price distribution should not be considered mere outliers, but relevant observations that reflect market segmentation and deserve to be analysed. For this reason, instead of adopting transformations that compress these values, it was considered more appropriate to apply quantile regressions, which make it possible to capture the effects of explanatory variables across the entire conditional distribution of prices, while avoiding parametric assumptions on the distribution of errors. In line with established practice in hedonic wine studies, the logarithmic transformation of the price was therefore adopted, as it ensures econometric robustness, guarantees comparability with the reference literature, and does not compromise the consistency of the estimators.

 

Table X. Comparison of price distributions (€/75cl)

 

Statistic	Price_75	ln_Price75	boxcox_Price75
Mean	19.10	2.75	1.19
Std. deviation	17.09	0.57	0.07
Minimum	4.90	1.59	0.95
1st quartile (Q25)	10.30	2.33	1.14
Median (Q50)	13.63	2.61	1.18
3rd quartile (Q75)	22.59	3.12	1.25
Maximum	187.88	5.24	1.37
Variance	292.18	0.32	0.01
Skewness	4.43	1.13	0.08
Kurtosis	30.87	4.54	2.58

Comment 4: Brand reputation beyond user reviews (e.g. by renowned and recognised wine critics) is of course not taken into account here.

Response 4: In our model we included two complementary dimensions of reputation. On the one hand, Score_A, Score_B, and Score_C reflect consumer ratings of the wineries (0–5), which we interpret as a proxy of brand reputation perceived on the market. On the other hand, the set of variables AW_ captures awards and ratings assigned by sectoral guides (e.g. Gambero Rosso, Bibenda, Vitae AIS, VITIS Vitae, Wine Spectator, Mundus Vini) and by internationally recognised wine critics (e.g. Parker, Maroni, Suckling). These refer to the evaluation of the specific wine labels, but also contribute to shaping the overall reputation of the winery. In this sense, our analysis takes into account both consumer-driven reputation signals and expert-driven quality recognitions, thus covering different facets of reputation in the wine market.

Comment 5: A cross-sectional study is, of course, only a snapshot and unfortunately does not allow any conclusions to be drawn about changes over time in the ‘willingness to pay’ for sustainability and identity practices.

Response 5: We fully agree with the reviewer. As noted in the limitations section (lines 608–612), the cross-sectional nature of the dataset does not allow us to capture changes over time in the perception of the analysed attributes. Considering that Primitivo is undergoing a phase of commercial repositioning, it would indeed be interesting to explore, in future research, whether and how the value attributed to identity-related or sustainable practices evolves over time.

Comment 5: The graphics in Figures 1 and 2 are legible, but unfortunately they are low resolution and axis labels are missing (units of the x and y axes in Figure 1 and units in Figure 2 on the x-axis). The legends are very concise, meaning that experts in the field will understand them, but new readers will need to familiarise themselves with them.

Response 5: We agree, in fact the figures have been eliminated.

Comment 6-7: I would have liked to read a little more detail about the software used, ‘STATA 15’ (command structures, setup, etc.).

When collecting data, I would also have liked to learn more about the source code of the scraper, or the libraries used (e.g. Python packages) as well as a description of any data cleansing that was carried out (if necessary).

Response 6-7: We have now specified in the methodological section that the data collection was carried out using Web Scraper, a browser-based tool for extracting structured data from e-commerce platforms, and the data processing took place through the use of STATA 15. We believe that this clarification improves the transparency of the data collection and processing process. As for the request for additional technical details (e.g. source code or libraries), we preferred to keep the description concise, in line with the format of the journal and with the attention paid to the noun results. However, we specify that standard data cleaning and validation procedures have been applied and that, on request, technical materials can be made available to interested readers.

Comment 8: The standard errors of the regression were estimated using ‘bootstrap’, but I could not find any information on the number of iterations, the confidence level or the selection of random samples, which at least makes reproducibility difficult.

Response 8: We specify that some information was already reported in the notes to the results tables, which read:

Notes: * p < 0.10; ** p < 0.05; p < 0.01.

^a Bootstrap standard errors in parentheses; Replications=1000.

^b Robust standard errors in parentheses; OLS = ordinary least squares, included as reference.

However, we acknowledge that other useful information was not specified. We have therefore added that the confidence level used is the default Stata (95%) and that, as far as the selection of random samples is concerned, the 54321 seed has been set to ensure full reproducibility of the results.

Comment 9: Of course, it would be wonderful if the analyses could be verified via supplementary data packages or an open-source repository for the data sets and analysis scripts. Perhaps you should include a reference stating that interested readers can obtain this from you on request?

Response 9: We agree on the importance of transparency and the possibility of replicating analyses. For this reason, we have added a note to the manuscript specifying that the datasets and scripts used for the analyses are available from the authors upon request.